-
Alma Mater Studiorum – Università di Bologna
DOTTORATO DI RICERCA IN
BIOLOGIA CELLULARE E MOLECOLARE
Ciclo XXVIII
Settore Scientifico Disciplinare: 05/E2
Settore Concorsuale di afferenza: BIO/11
Structural characterization of
meningococcal vaccine antigen NadA
and of its transcriptional regulator NadR
in ligand-bound and free forms.
Presentato da: Alessia Liguori
Coordinatore Dottorato
Chiar.mo Prof. Davide Zannoni
Relatore
Dr. Matthew James Bottomley
Co-relatore
Dr. Enrico Malito
Co-relatore
Chiar.mo Prof. Vincenzo Scarlato
Esame finale anno 2016
-
2
-
3
Contents
Aim of the thesis
................................................................................................................
6
Introduction
........................................................................................................................
8
Reverse Vaccinology and Structural
Vaccinology................................................................
9
The Serogroup B Meningococcus
Vaccine..........................................................................
11
Neisseria meningitidis colonization and
invasion…..……….................................................
12
Meningococcal virulence factors and
adhesion....................................................................
13
Transcriptional
regulators.....................................................................................................
14
Genome plasticity and phase
variation.................................................................................
15
Part One
Crystal structures reveal the molecular basis of
ligand-dependent regulation of NadR,
the transcriptional repressor of the meningococcal antigen
NadA............................... 17
Abstract...............................................................................................................................
18
The Neisserial adhesin Regulator
(NadR)............................................................................
19
The MarR family of transcriptional
regulators.......................................................................
19
Experimental Procedures
.................................................................................................
21
Bacterial strains, culture conditions and mutant
generation................................................. 21
Protein production and
purification.......................................................................................
22
Size-exclusion high-performance liquid chromatography (SE-HPLC)
coupled with
Multi-angle laser light scattering
(MALLS)............................................................................
22
Differential Scanning Calorimetry
(DSC)..............................................................................
23
Surface plasmon resonance
(SPR)......................................................................................
23
Crystallization of NadR in the presence or absence of
4-HPA............................................. 24
X-ray diffraction data collection and structure
determination................................................
24
Results
................................................................................................................................
26
NadR is dimeric and is stabilized by specific
hydroxyphenylacetate ligands........................ 26
NadR displays distinct binding affinities for
hydroxyphenylacetate ligands.......................... 28
Crystal structures of holo-NadR and
apo-NadR...................................................................
29
The holo-NadR structure presents only one occupied
ligand-binding pocket....................... 31
-
4
Analysis of the pockets reveals the molecular basis for
asymmetry and stoichiometry…... 32
Apo-NadR structures reveal conformational
flexibility..........................................................
34
4-HPA stabilizes concerted conformational changes in NadR that
prevent DNA-binding… 35
A single conserved leucine residue (L130) is crucial for
dimerization.................................. 37
Discussion
..........................................................................................................................
40
Part Two
New strategies towards a crystal structure of meningococcal
antigen NadAv3.......... 47
Abstract................................................................................................................................
48
The Neisseria meningitidis adhesin A
(NadA).......................................................................
49
The Trimeric Autotransporter Adhesin (TAA) family – structural
organization...................... 49
NadA
variants……………………………................................................................................
51
Experimental Procedures
..................................................................................................
52
NadA constructs cloning, expression and
mutagenesis........................................................
52
NadA protein production and
purification…………………..……………….............................. 52
Size-exclusion high-performance liquid chromatography (SE-HPLC)
coupled with
Multi-angle laser light scattering
(MALLS).............................................................................
52
Differential Scanning Calorimetry
(DSC)...............................................................................
53
Surface plasmon resonance
(SPR).......................................................................................
53
Crystallization of NadAv3 proteins and X-ray diffraction data
collection............................... 54
Results
................................................................................................................................
56
Structure based-design: seeking the minimal N-terminal domain of
NadAv3 ………........... 56
Strategies towards the crystal structure of NadAv324-170
construct........................................ 58
NadAv3 mutagenesis to promote
crystallization...................................................................
62
High-quality NadA reagents for functional
studies................................................................
65
Discussion…………….........................................................................................................
67
Acknowledgments……………............................................................................................
71
References…………….........................................................................................................
72
-
5
“A Matthew, guida e stimolo
per la mia crescita professionale e personale.”
-
6
Aim of the thesis
Serogroup B Neisseria meningitidis (MenB) is the cause of an
acute, potentially
severe infection, known as invasive meningococcal disease (IMD)
with two peaks in disease
incidence occurring among adolescents and young adults 16 to 21
years of age. Bexsero is
the first genome-derived vaccine against MenB, and it has
recently been approved in >35
countries worldwide. Neisserial adhesin A (NadA), a
meningococcal trimeric autotransporter
adhesin (TAA) that acts in adhesion to, and invasion of, host
epithelial cells, is one of the
three protein antigens included in Bexsero. The main aim of this
work was to obtain detailed
insights into the structure of NadA variant 3 (NadAv3), the
vaccine variant, and into the
molecular mechanisms governing its transcriptional regulation by
NadR (Neisseria adhesin A
Regulator). The amount of NadA exposed on the meningococcal
surface influences the
antibody-mediated serum bactericidal response measured in vitro,
which in turn correlates
with protection in immunized subjects. A deep understanding of
nadA expression is therefore
important, otherwise the contribution of NadA to vaccine-induced
protection against
meningococcal disease may be underestimated. The abundance of
surface-exposed NadA is
regulated by the ligand-responsive transcriptional repressor
NadR. The functional,
biochemical and high-resolution structural characterization of
NadR is presented in the first
part of the thesis (Part One). These studies provide detailed
insights into how small molecule
ligands, such as hydroxyphenylacetate derivatives, found in
relevant host niches, modulate
the structure and activity of NadR, by ‘conformational
selection’ of inactive forms. These
findings shed light on the regulation of a key virulence factor
and vaccine antigen of this
important human pathogen.
In the second part of the thesis (Part Two), strategies
involving both protein
engineering and crystal manipulation to increase the likelihood
of solving the crystal structure
of NadAv3 are described. The first approach was the rational
design of new constructs of
NadAv3, based on the recently solved crystal structure of a
close sequence variant
(NadAv5). Then, a comprehensive set of biochemical, biophysical
and structural techniques
were applied to investigate all the generated NadAv3 constructs,
aiming to faithfully
represent its natural trimeric status, essential for reliable
structural, functional and epitope
mapping studies. The well-characterized trimeric NadAv3
constructs represented a set of
high quality reagents which were validated as probes for
functional studies and as a platform
for continued attempts for protein crystallization. Mutagenesis
studies and screenings to
identify a new crystal form of NadAv3 were performed to improve
crystal quality, ultimately
allowing the collection of several high quality X-ray
diffraction data sets; structure
determination is ongoing. The atomic resolution structure of
NadAv3 will help to understand
its biological role as both an adhesin and a vaccine antigen.
For example, the high resolution
structure will enable epitope mapping studies using human
antibodies and thus permit a
-
7
deeper understanding of the molecular determinants of antibody
binding and protective
epitopes. In addition, it will help to understand the molecular
basis of host-pathogen
interactions mediated by specific human cell receptors.
-
8
Introduction
-
9
Reverse Vaccinology and Structural Vaccinology
Genomics tools and the exponentially growing number of bacterial
genome
sequencing projects have changed the landscape of modern biology
providing new
opportunities for vaccine development. The complete genome of a
bacterium represents a
large reservoir of genes encoding for potential antigens that
can be selected and tested as
vaccine candidates. Therefore, potentially surface-exposed
proteins can be identified in a
reverse manner, starting from the genome rather than from the
microorganism. This
approach has been termed Reverse Vaccinology (RV) [1].
Bioinformatics algorithms are
used to select open reading frames (ORFs) encoding putative
surface-exposed or secreted
proteins, which are potentially recognized by antibody and can
therefore considered as
vaccine antigens. The identification of such surface proteins is
based on specific properties
including the presence of signal peptide sequences, membrane
spanning regions, lipoprotein
signature, and motifs such as LPXTG sortase attachment sites.
Sequence homology
analyses can additionally help the antigen identification
process, comparing homology both
to known virulence factors or protective antigens from other
pathogens and to human
proteins to avoid autoimmune problems [2]. The candidate surface
antigens are therefore
produced as recombinant proteins and tested for their
immunogenicity in a relevant animal
model in order to evaluate their potential as vaccine
candidates. The reverse vaccinology
approach has been strengthened by the development of proteomic
techniques to identify
vaccine candidates against bacterial infections [3-5]. In the
proteomics approach to bacterial
vaccine development the surface-located or secreted bacterial
proteins are first separated
using two-dimensional (2-D) electrophoresis gel, followed by
digestion of each protein into its
peptide fragments using a specific protease (e.g. trypsin). The
molecular mass of each
proteolytic digested fragment is then accurately measured using
matrix-assisted laser
desorption/ionization time-of-flight mass spectrometry
(MALDI-TOF MS) [6, 7]. Finally, the
generated peptide mass fingerprint is used as an input to allow
a database search of
predicted masses coming from the digestion of a list of known
proteins. If a protein sequence
in the reference list does not match the experimental values,
the peptide can be identified
using tandem mass spectrometry, which provides sequence
information on the proteolytic
peptides [8]. The effectiveness of the 2-D gel-based platform
integrating surface and
immune-proteomics analysis was demonstrated by the
identification of major meningococcal
vaccine antigens [9]. After identification of potential
candidates by RV, their testing is
facilitated by use of high-throughput screenings for protective
immunity and correlation of
protection. Combining proteomics with serological analysis is
another useful refinement for
identifying potential vaccine candidates [8].
In addition, systematic transcriptomic and proteomic gene
expression analysis can
support RV in the identification of gene-level responses, which
are correlated with protection
-
10
in vivo and then facilitating the rational design of a
hypothetical vaccine candidate [10].
These strategies could help vaccine design for pathogens for
which vaccines are not yet
available, as well as parasites and viruses.
Once candidates have been identified further potential issues
have to be solved
before they can be used in a vaccine formulation. For instance,
the surface antigens
identified are not always abundant, and immune evasion
strategies set up by the micro-
organism can impact their potential. One of the most frequent
examples is the sequence
variation of surface antigens across circulating strains. From a
more practical viewpoint, the
selected antigens may show low stability when expressed as
recombinant proteins.
Structural vaccinology (SV) can represent the solution for many
of these issues [11]. In a
process analogous to structure-based drug design of
small-molecule pharmaceuticals, where
lead candidate inhibitory molecules are rationally-optimized in
structure-guided manner,
structural information on antigens and their protective epitopes
can also be instrumental
during the optimization phases.
Notable applications of structural vaccinology in the field of
bacterial protein antigens
include (i) characterization of the immune response through
epitope mapping to provide
insights into the molecular features recognized by the host
immune response upon infection
by the pathogen or following immunization. Epitope mapping
experiments produce
information about the immunogenic regions of the protein,
showing which parts of a surface-
bound antigen are exposed and therefore accessible to
antibodies. Another important
application for structural vaccinology is (ii) the possibility
to improve the biochemical stability,
and homogeneity of a candidate, stabilizing the folding and
reducing degradation and
tendency to aggregate. It is also possible (iii) to engineer a
protein antigen in order to
overcome limits imposed by sequence variability. SV can drive
the design of chimeric
antigens that display epitopes from multiple proteins to elicit
an immune response with wider
specificity. Overall, a structurally re-designed molecule can
become an antigen with
increased immunogenicity, and efficacy. A successful SV approach
will also facilitate scale
up, and generate an antigen that can be more easily produced,
more homogeneous and
stable over time. In summary, SV can use knowledge of
biochemical, biophysical, structural,
immunological & functional properties of biomolecules to
benefit vaccine development,
encompassing several steps of the process starting from antigen
selection up to vaccine
approval.
An example of how a structure-based approach has already been
used in a
preclinical vaccine design program is provided by a combination
of NMR spectroscopy and
X-ray crystallography to obtain structural insights of the
immunodominant domain of
GNA1870, a protective antigen of N. meningitidis identified by
RV. The epitopes of
-
11
bactericidal antibodies against several meningococcal strain
variants were mapped onto the
NMR structure of GNA1870, providing the basis for the rational
design of an engineered form
of GNA1870 containing several cross-protective, B cell epitopes.
A protein domain is defined
as an independent unit that can have an independent function in
a single-domain protein or
can contribute to the function of a multidomain protein in
cooperation with other domains
[12]. The new GNA1870 antigen had a conserved backbone that
carried an engineered
surface containing specificities for all three variant groups,
demonstrating that the structure-
based design of an engineered antigen is an efficient way to
generate a broadly protective
antigen [13].
The Serogroup B Meningococcus Vaccine
The concept of reverse vaccinology was developed and applied for
the first time to
N. meningitidis serogroup B (MenB). N. meningitidis is the major
cause of meningitis and
sepsis, two devastating diseases that can kill children and
young adults within hours, despite
the availability of effective antibiotics. N. meningitidis is a
Gram-negative bacterium that
colonizes asymptomatically the upper nasopharynx of about 5–15%
of the human population,
establishing a commensal relationship between the host and the
bacterium that fails or
becomes dysfunctional in case of disease [14]. This condition
represents the only known
reservoir for meningococcal infection but may also contribute to
establishing host immunity
[15]. For unknown factors dependent on both the host and
pathogen, the meningococcus can
invade the pharyngeal mucosal epithelium and disseminate into
the bloodstream causing
septicaemia or cross the blood-brain barrier and enter the
cerebrospinal fluid, causing
meningitis. Although reasons leading to the bacterial invasion
are not well known,
environmental factors that damage the nasopharyngeal mucosa,
together with the lack of a
protective immune response could increase the incidence of
invasive meningococcal
disease. N. meningitidis can be classified in 13 serogroups on
the basis of the chemical
composition of the capsule polysaccharide, five of which (A, B,
C, W-135 and Y) are
responsible for more than 95 % of total cases of invasive
disease. Vaccines against
serogroups A, C, W-135 and Y were developed in the 1960s by
using the purified capsular
polysaccharide as antigen. At the turn of the century, improved
second-generation,
conjugated vaccines were introduced, where the polysaccharide
components were linked to
a carrier protein, and which provide effective protection in all
age groups [16]. However, the
chemical composition of the polysaccharide of serogroup B, which
resembles a molecule
present in human tissues, makes a polysaccharide-based vaccine
poorly immunogenic and a
possible cause of autoimmunity. In the last 40 years much effort
has been directed to the
identification of meningococcus B protein antigens as the basis
of new vaccines. However,
-
12
the high variability of these proteins among the different MenB
strains represents a serious
obstacle to the production of a globally effective anti-MenB
vaccine [15]. Reverse
vaccinology has therefore proven to be a rapid and reliable
approach to identifying vaccine
candidates. The three most immunogenic antigens on the basis of
their ability to induce
bactericidal activity or in vivo passive protection were
selected to be used in a
multicomponent vaccine. They were NHBA [17], fHbp [18, 19], and
NadA [20, 21]. Two other
antigens (named GNA2091 and GNA1030) were also selected. To
further enhance their
immunogenicity and facilitate large-scale manufacturing of the
vaccine components, four of
the selected antigens were combined into two fusion proteins so
that the resulting protein
vaccine contained three recombinant proteins. The antigen NHBA
was fused to GNA1030
while GNA2091 was fused to fHbp. NadA was included as a single
antigen as it did not
perform well when fused to a partner [22-24].
N. meningitidis colonization and invasion
Colonization of the upper respiratory mucosal surfaces by N.
meningitidis is the first
step in the establishment of a human carrier state and invasive
meningococcal disease
(Figure 1).
Figure 1. Stages in the pathogenesis of meningococcus. N.
meningitidis may be acquired through the inhalation of respiratory
droplets. The organism establishes intimate contact with
non-ciliated mucosal epithelial cells of the upper respiratory
tract, where it may enter the cells briefly before migrating back
to the apical surfaces of the cells for transmission to a new host.
Asymptomatic carriage is common in healthy adults in which bacteria
that enter the body by crossing the epithelial barrier are
eliminated. In susceptible individuals, once inside the blood, N.
meningitidis may survive, multiply rapidly and disseminate
throughout the body and the brain. Meningococcal passage across the
brain vascular endothelium (or the epithelium of the choroid
plexus) may then occur, resulting in infection of the meninges and
the cerebrospinal fluid [25, 26].
-
13
Initial contact of meningococci with nasopharyngeal epithelial
cells is mediated by
Type IV pili, the host receptor for which may be the I-domain of
integrin α chains or possibly
CD46 [27]. At this level, the downregulation of the capsule may
activate attachment [26],
allowing meningococci a closer adherence to the host epithelial
cells. It results in the
formation of cortical plaques and leads to the recruitment of
factors ultimately responsible for
the formation and extension of epithelial cell pseudopodia that
engulf the meningococcus.
Intimate association is mediated by interaction of the bacterial
opacity proteins, Opa and
Opc, with CD66/CEACAMs and integrins, respectively, on the
surface of the epithelial cell
and is one trigger of meningococcal internalization [28]. The
next steps of meningococcal
internalization, intracellular survival and transcytosis through
the basolateral tissues and
dissemination into the bloodstream are less well studied
[15].
Meningococcal virulence factors and adhesins
The virulence of N. meningitidis is influenced by multiple
factors, including both
genetic mechanisms, allowing the bacteria to vary its phenotype
and adapt to the host, and
iron sequestration mechanisms. Additionally, meningococci
express multiple molecules
acting as endotoxin, secreted factors or surface proteins,
located in different compartments
of the menincococcal cell membrane, which interact with host
cellular molecules. The key
structures at the interface between the meningococcus and the
host are the polysaccharide
capsule and/or lipopolysaccharide (LPS) that may shield
bacterial surfaces from the host
innate and adaptive immune effector mechanisms, and the
protruding surface proteins that
are known as pili [26]. Pili are filamentous structures
consisting of protein subunits that
extend from the bacterial surface, and these seem to be the main
players in the initiation of
the interaction between meningococcus and the host cell [29,
30]. Pili facilitate adhesion to
host tissues, further aided by the outer membrane adhesins Opa
and Opc. The opacity
proteins (Opa and Opc) are integral outer membrane proteins that
mediate pathogen-host
interaction adhering to and invading epithelial and endothelial
cells. Both bind the heparan
sulphate proteoglycans and sialic acids [31, 32] but they also
display a degree of receptor
specificity [32].
Numerous additional minor adhesins are generally expressed at
low levels during in
vitro growth but may be important in in vivo infections.
Neisseria hia homologue A (NhhA)
mediates low levels of adhesion to epithelial cells, to heparan
sulphate proteoglycans
(HspGs) and to laminin [33]. Adhesion penetration protein (App)
regulates interactions
between the bacteria and the host tissue by mediating adhesion
during the early stages of
colonization, before it is autocleaved. At later stages, App
autocleavage may allow bacterial
detachment, therefore facilitating bacterial spread [34].
Meningococcal serine protease A
-
14
(MspA) expressed by several but not all virulent Neisseria
strains mediates binding to both
epithelial and endothelial cells and to elicit the production of
bactericidal antibodies [35].
Multiple adhesin family (Maf) is a family of glycolipid adhesins
characterised first for the
gonococcus that may play a role in Opa-independent invasion
[36]. Neisserial adhesin A
(NadA) is a member of the trimeric autotransporter adhesins
(TAAs) belonging to the Oca
(oligomeric coiled-coil adhesin) family and is involved in
adhesion and invasion of N.
meningitidis [20, 21].
The regulation of NadA is part of this study and is discussed in
Part One of this
thesis, while the strategies towards a crystal structure of the
NadA variant 3 will be described
in Part Two.
Transcriptional regulators
During infection, N. meningitdis can invade diverse sites within
the human host, which
represent different niches with respect to nutrients,
environmental factors and competing
microorganisms. Therefore it is subjected to constant selective
pressures, and its ability to
rapidly adapt its metabolism and cellular composition to
environmental changes is essential
for its survival [37]. Bacteria have two major and complementary
mechanisms for adapting to
changes in their environment: changing their genotype (genome
plasticity) or altering gene
expression, both leading to phenotypic variations. The
differential expression of potential
virulence factors depends largely on the activity of
transcriptional regulators, whose activity
plays an important role for example in the infection process of
N. meningitidis. Relatively few
transcriptional regulators are found in the pathogenic
Neisseriae [38]; 36 putative regulators
in N. meningitidis (strain MC58) and 34 in N. gonorrhoeae
(strain FA1090), compared to the
free-living E. coli, which harbours more than 200
transcriptional regulators. The paucity in
transcriptional regulators may possibly be related to the
restricted ecological niche of the
Neisseria spp. which are human-adapted pathogens for which there
is no other known
reservoir. Until now, only few of the predicted 36
transcriptional regulators in N. meningitidis
MC58 have been characterized. Two of the transcriptional
regulators in N. meningitidis are
members of the MarR family and are encoded by the genes NMB1585
and NMB1843. The
structure of the transcription factor NMB1585 has been solved,
but its physiological role has
not been characterized and therefore the identity of any natural
ligand(s) that may modulate
its activity is unknown [39]. The product of the NMB1843 gene is
the Neisserial adhesin
Regulator (NadR) - a MarR-family transcriptional regulator of
16.6 kDa per monomer and it
has been demonstrated to repress expression of the meningococcal
adhesin NadA [40, 41].
NadR is the subject of an extensive structural and biophysical
characterization in this thesis
and will be discussed below in detail (Part One).
-
15
Several MarR-family transcriptional regulators have previously
been identified and
described for their activity. In meningococcus: the ferric
uptake regulator (Fur) is involved in
the response to iron [42-44] and has even been shown to
indirectly control gene expression
through small regulatory RNA molecules [40, 45]; Zur is the
second Fur-like regulator that
responds specifically to Zn2+ and controls Zn2+- uptake by
regulating a TonB receptor that
functions in high affinity Zn2+ acquisition [46]. Adaptation to
oxygen-limited conditions as
encountered during infection of the human host is mediated by
the transcriptional activator
FNR (Fumarate and Nitrate Reductase regulator), whose
DNA-binding ability is stabilized in
the presence of oxygen [47, 48]. Upon conditions of oxygen
limitation, this regulator enables
the meningococci to survive by switching to enhanced sugar
fermentation and expression of
a denitrification pathway, utilizing nitrite instead of oxygen
as a respiratory substrate [49].
NsrR acts as a repressor of a regulon of genes which responds to
nitric oxide [50, 51]. The
LysR-type regulator CrgA is upregulated upon contact with human
epithelial cells [52]; it acts
as a repressor of transcription of its own gene and as an
activator of transcription of the
mdaB gene [53]. NMB0573 (annotated as AsnC) is a global
regulator controlling the
response to poor nutrient conditions, which are perceived by
binding of this regulator to
leucine and methionine, two amino acids representing general
nutrient abundance [54].
Although extensive transcriptional regulation is expected to
accompany both the
survival and the infection process of N. meningitidis, limited
information about transcriptional
regulation is available. Only a few of the predicted regulators
have been characterized and
the regulons of even fewer have been deeply studied, including
those involved in the
adaptation of meningococcus to iron and oxygen limitation and
response to nitric oxide.
Genome plasticity and Phase variation
In order to adapt to changing microenvironments and avoid the
host immune
defences, the meningococcus possesses mechanisms for rapid
genome variation and
diversification. The genome plasticity is promoted by
spontaneous mutational mechanisms.
These events originate either from local genomic changes caused
by repeat sequences,
phase and antigenic variation, recombination and horizontal gene
transfer, or globally from
mutated alleles. Repeat sequence elements facilitate the
duplication or deletion of regions of
the genome, as well as recombination, and thereby establish
small and large alterations. The
addition and deletion of repeat units lead to the molecular
mechanism of phase variation in
Neisseria, most often owing to slipped-strand mispairing (SSM).
The presence of repeat units
causes a slippage of the synthesis strand over the template
strand during replication that
leads to the addition or the deletion of units in the newly
synthesised strand [55]. The number
of repeats can influence translation or transcription by
introducing frameshift mutations or
-
16
changing critical promoter spacing, resulting in high frequency
on-off switching or modulation
of the level of expression of genes usually associated with
surface-exposed antigens [56-59].
In meningococcus a considerably high quantity of phase-variable
genes have been identified
in which phase variation is used to alter surface-exposed
molecules such as outer-
membrane proteins PorA, Opc, Opa, pili and specific adhesins, as
well as LPS and capsule
[55, 60, 61]. In particular, the expression of NadA is phase
variable and a tetranucleotide
tract (TAAA) located upstream of the nadA gene promoter has been
demonstrated to control
this phenomenon, through an altered sigma-factor binding [40].
Whole-genome-sequence
analyses have largely confirmed the importance of varying
surface-exposed antigens for
allowing bacterial commensals and pathogens to evade the immune
system of their host and
to adapt to changeable environments.
Concluding remarks
Host-pathogen interaction is a dynamic process that can lead to
different outcomes
such as an equilibrium known as commensalism or the
establishment of a disease. The
factors that lead N. meningitidis to establish the infection,
switching from commensal to
pathogenic are still poorly understood. For these reasons, a
better understanding of the
causes and mechanism that mediate the expression of proteins
involved in the interaction
with host tissues is needed, both for predicting the
effectiveness of a vaccine which contains
these proteins and for characterizing at the molecular level
novel strategies of bacterial
populations to changing host environments. N. meningitidis has
to change gene expression
repertoire in order to adapt and survive in the different
tissues during an infection of the
human host. At the same time the bacteria escape the host immune
response, which targets
mostly the same structures used by the meningococcus to interact
with the host, by surface
structure expression variability and redundancy. A
multi-disciplinary approach based on
molecular genetics, biochemical, biophysical and structural
analyses, will provide molecular
knowledge of the transcriptional regulation of antigen
expression.
-
17
Part One
Crystal structures reveal the molecular basis of
ligand-dependent regulation of NadR,
the transcriptional repressor of the meningococcal
antigen NadA
-
18
Abstract
Neisseria adhesin A (NadA) is present on the meningococcal
surface and contributes
to adhesion to and invasion of human cells. NadA is also one of
three recombinant antigens
in the Bexsero vaccine, approved in 2012 by the European
Medicines Agency (EMA), which
protects against serogroup B meningococcus. The amount of NadA
on the bacterial surface
influences the antibody-mediated serum bactericidal response
measured in vitro. It is
therefore important to understand the mechanisms which regulate
nadA expression levels,
which are predominantly controlled by the transcriptional
regulator NadR (Neisseria adhesin
A Regulator) both in vitro and in vivo, otherwise the real
contribution of NadA to vaccine-
induced protection against meningococcal meningitis may be
underestimated. NadR binds
the nadA promoter and represses gene transcription. In the
presence of 4-
hydroxyphenylacetate (4-HPA), a catabolite present in human
saliva both under physiological
conditions and during bacterial infection, the binding of NadR
to the nadA promoter is
attenuated and nadA expression is induced. NadR also mediates
ligand-dependent
regulation of many other meningococcal genes, for example the
highly-conserved multiple
adhesin family (maf) genes, which encode proteins emerging with
important roles in host-
pathogen interactions, immune evasion and niche adaptation. To
gain insights into the
regulation of NadR mediated by 4-HPA, the work presented here
combined X-ray
crystallographic, biochemical, and mutagenesis studies. In
particular, two new crystal
structures of ligand-free and ligand-bound NadR revealed (i) the
molecular basis of
‘conformational selection’ by which one molecule of 4-HPA binds
and stabilizes dimeric
NadR in a conformation apparently unsuitable for DNA-binding,
(ii) molecular explanations
for the binding specificities of different hydroxyphenylacetate
ligands, including 3Cl,4-HPA
which is produced during inflammation, (iii) the presence of a
leucine residue essential for
dimerization and conserved in many MarR family proteins, and
(iv) four residues (His7, Ser9,
Asn11 and Phe25), which are involved in binding 4-HPA, and were
confirmed in vitro to have
key roles in the regulatory mechanism in bacteria. Overall, this
study deepens our molecular
understanding of the sophisticated regulatory mechanisms of the
expression of nadA and
other genes governed by NadR, dependent on interactions with
niche-specific signal
molecules that may play important roles during meningococcal
pathogenesis.
-
19
The Neisserial adhesin Regulator (NadR)
Previous studies revealed that nadA expression levels are mainly
regulated by the
Neisseria adhesin A Regulator (NadR) [41]. Although additional
factors influence nadA
expression, the attention was focused on the regulation by NadR,
the major mediator of
nadA phase variable expression [62, 63]. Studies of NadR also
have broader implications,
since a genome-wide analysis of MenB wild-type and nadR
knock-out strains revealed that
NadR influences the regulation of >30 genes, including maf
genes, from the multiple adhesin
family [64]. These genes encode a wide variety of proteins
connected to many biological
processes contributing to bacterial survival, adaptation in the
host niche, colonization and
invasion [65, 66]. NadR binds the nadA promoter and represses
gene transcription [63].
NadR binds nadA on three different operators (OpI, OpII and
OpIII) [64]. The DNA-binding
activity of NadR is attenuated in vitro upon addition of various
hydroxyphenylacetate (HPA)
derivatives, including 4-HPA. 4-HPA is a small molecule derived
from mammalian aromatic
amino acid catabolism and released in human saliva, where it has
been detected at
micromolar concentration [67]. In the presence of 4-HPA, NadR is
unable to bind the nadA
promoter and nadA gene expression is induced [63, 64]. In vivo,
the presence of 4-HPA in
the host niche of N. meningitidis serves as an inducer of NadA
production, thereby promoting
bacterial adhesion to host cells [64]. Further, it was recently
reported that 3Cl,4-HPA,
produced during inflammation, is another inducer of nadA
expression [68]. However, the
molecular mechanism explaining how this transcriptional
regulator interacts with 4-HPA or its
derivatives and modulates their DNA-binding affinities
accordingly has remained unresolved.
The structural analysis of NadR was attempted in order to
illustrate precisely how this protein
recognizes and binds 4-HPA and to provide the structural basis
for the attenuated DNA
binding of NadR upon its interaction with 4-HPA.
The MarR family of transcriptional regulators
NadR belongs to the MarR (Multiple Antibiotic Resistance
Regulator) family, a group
of ligand-responsive transcriptional regulators ubiquitous in
bacteria and archaea. MarR
family proteins can promote survival in the presence of
antibiotics, toxic chemicals, organic
solvents or reactive oxygen species [69, 70] and can regulate
virulence factor expression
[71]. MarR homologues can act either as transcriptional
repressors or as activators [72]. To
date, >50 MarR family structures are known, revealing a
conserved fold of six α-helices (H)
and a two-stranded antiparallel -sheet (B) in the topology:
H1-H2-H3-H4-B1-B2-H5-H6. The
DNA-binding domains are ascribed to the superfamily of winged
helix proteins, containing α-
helices 3 and 4, comprise the helix-turn-helix motif, and the
-sheet is called the wing. Helix
4 is termed the recognition helix, as in other HTHs where it
binds the DNA major groove. The
-
20
α-helices 1, 5 and 6 are involved in dimerisation, as most
MarR-like transcription regulators
form dimers. Further, a few examples have been obtained in
complexes with target DNA
ligands. For example, the structure of the Bacillus subtilis
OhrR-ohrA complex revealed the
chimeric nature of the wHTH motif and a double-helix DNA binding
element, both of which
are proposed to be utilized by the entire MarR family to bind
cognate DNA [73]. A molecular
understanding of their ligand-dependent regulatory mechanisms is
still limited, often
hampered by lack of identification of their ligands. A
potentially interesting exception comes
from the ligand-free and salicylate-bound forms of the
Methanobacterium
thermoautotrophicum protein MTH313 which revealed that two
salicylate molecules bind to
one MTH313 dimer and induce large conformational changes,
apparently sufficient to
prevent DNA binding [74]. However, the homologous archeal
Sulfolobus tokodaii protein
ST1710 presented essentially the same structure in ligand-free
and salicylate-bound forms,
apparently contrasting the mechanism proposed for MTH313 [75].
Despite these apparent
differences, MTH313 and ST1710 bind salicylate in approximately
the same site, between
their dimerization and DNA-binding domains. However, it is
unknown whether salicylate is a
relevant in vivo ligand of either of these two proteins, which
share ~20% sequence identity
with NadR, rendering unclear the interpretation of these
findings regarding the regulatory
mechanisms of NadR or other MarR family proteins [72]. Other two
MarR family homologues
TcaR and SAR2349 from Staphylococcus epidermidis and
Staphylococcus aureus,
respectively, have been crystallized in the presence of
salicylate and antibiotics. In the
structure of TcaR complexed with salicylate, multiple binding
site where found, one of which
(SAL-1) overlaps with the binding site seen in MTH313 [76]. The
structures of SAR2349–
antibiotic complexes reveals that the binding of antibiotics
change the angle between the
dimerization domains, inducing conformational changes within the
wHTH motifs that
interferes with binding to DNA [77].
-
21
Experimental procedures
Bacterial strains, culture conditions and mutant generation. In
this study
N. meningitidis MC58 wild type strain and related mutant
derivatives were used. The MC58
isolate was kindly provided by Professor E. Richard Moxon,
University of Oxford, UK, and
was previously submitted to the Meningococcal Reference
Laboratory, Manchester, UK [78].
Strains were routinely cultured, stocked, and transformed as
previously described [64]. The
preparation of the expression construct enabling production of
soluble NadR (Uniprot code
Q7DD70) with an N-terminal His-tag followed by a thrombin
cleavage site and NadR
residues M1-S146 was described previously [79]. Site-directed
mutagenesis was performed
using two 2 couples of mutagenic primers containing the desired
mutation to amplify pET15b
containing several NMB1843 variants. In short, 1 to 10 ng of
plasmid template were amplified
using Kapa HiFi DNA polymerase (Kapa Biosystems) and the
following cycling conditions:
98°C for 5 min, 15 amplification cycles (of 98°C for 30 s, 60°C
for 30 s, 72°C for 6 min)
followed by a final extension of 10 min at 72°C. Residual
template DNA was digested by
30 min incubation with FastDigest DpnI (Thermo Scientific) at
37°C and 1 µl of this reaction
was used for transforming competent E. coli DH5α.
Strains or plasmid Relevant characteristics
E. coli strains
DH5α supE44 lacU169 (w80lacZDM15) hsdR17 recA1 endA1 gyrA96
thi-1 relA1
Plasmids
pET15b-1843 pET15b derivative for expression of recombinant
NMB1843, AmpR
pET15b-PDD0
pET15b derivative for expression of recombinant NMB1843
containing an H7A mutation, AmpR
pET15b-PDD1
pET15b derivative for expression of recombinant NMB1843
containing an S9A mutation, AmpR
pET15b-PDD2
pET15b derivative for expression of recombinant NMB1843
containing an N11A mutation, AmpR
pET15b-PDD3
pET15b derivative for expression of recombinant NMB1843
containing an Y115A mutation, AmpR
pET15b-PDD4
pET15b derivative for expression of recombinant NMB1843
containing an K126A mutation, AmpR
pET15b-PDD5
pET15b derivative for expression of recombinant NMB1843
containing L130K and L133K mutations, AmpR
pET15b-PDD6
pET15b derivative for expression of recombinant NMB1843
containing K126A, L130K and L133K mutations, AmpR
pET15b-PDD7
pET15b derivative for expression of recombinant NMB1843
containing N11A, D112A, R114A and Y115A mutations, AmpR
pET15b-PDD8
pET15b derivative for expression of recombinant NMB1843
containing an L130K mutation, AmpR
pET15b-PDD9
pET15b derivative for expression of recombinant NMB1843
containing an L133K mutation, AmpR
-
22
Protein production and purification. The NadR expression
constructs (wild-type or
mutant clones) were transformed into E. coli BL21 (DE3) cells
and were grown as 500mL
culture volumes in 2L shake flasks at 37°C in Luria-Bertani (LB)
medium supplemented with
100µg/mL ampicillin, until an OD600 of 0.5 was reached. Target
protein production was
induced by the addition of 1mM IPTG followed by incubation with
shaking overnight at 21°C.
(For production of the SeMet derivative form of NadR,
essentially the same procedure was
followed, but using the E. coli B834 strain grown in a modified
M9 minimal medium
supplemented with 40mg/L L-selenomethionine). Cells were
harvested by centrifugation
(6400g, 30 min, 4°C), resuspended in 20mM HEPES pH 8.0, 300mM
NaCl, 20mM imidazole,
and were lysed by sonication (Qsonica Q700). Cell lysates were
clarified by centrifugation at
2800g for 30 min, and the supernatant was filtered using a
0.22µm membrane (Corning filter
system) prior to protein purification.
NadR was purified by affinity chromatography using an AKTA
purifier (GE
Healthcare). All steps were performed at room temperature
(18-26°C), unless stated
otherwise. The filtered supernatant was loaded onto an Ni-NTA
resin (5mL column, GE
Healthcare), and NadR was eluted using 4 steps of imidazole at
20, 30, 50 and 250mM
concentration, at a flow rate of 5mL/min. Eluted fractions were
examined by reducing and
denaturing SDS-PAGE analysis. Fractions containing NadR were
identified by a band
migrating at ~17kDa, and were pooled. The N-terminal 6-His tag
was removed enzymatically
using the Thrombin CleanCleave Kit (Sigma-Aldrich).
Subsequently, the sample was
reloaded on the Ni-NTA resin to capture the free His tag (or
unprocessed tagged protein),
thus allowing elution in the column flow-through of tagless NadR
protein, which was used in
all subsequent studies. The NadR sample was concentrated and
loaded onto a HiLoad
Superdex 75 (16/60) preparative size-exclusion chromatography
(SEC) column equilibrated
in buffer containing 20mM HEPES pH 8.0, 150mM NaCl, at a
flow-rate of 1mL/min. NadR
protein was collected and the final yield of purified protein
obtained from 0.5L growth medium
was approximately 8mg (~2mg protein per g wet biomass). Samples
were used immediately
for crystallization or analytical experiments, or were frozen
for storage at -20°C.
Size-exclusion high-performance liquid chromatography (SE-HPLC)
coupled
with Multi-angle laser light scattering (MALLS). SE-HPLC was
used to assess the purity
and the apparent molecular weight of the recombinant wild-type
NadR sample alone or
containing a 200-fold molar excess of 4-HPA and of the mutated
NadR samples. SE-HPLC
experiments were performed by loading 20μl of each sample at a
concentration of ~ 50μM on
an analytical size exclusion TSK Super SW3000 column of 4 μm
particle size and 250Å pore
size (Tosoh), with a separation range suitable for globular
proteins of 10 to 500 kDa.
Samples were eluted isocratically in 0.1M NaH2PO4, 0.4M
(NH4)2SO4 buffer at pH 6.0,
experiments were performed at room temperature (18-26°C).
-
23
MALLS analyses were performed in order to determine the absolute
molecular mass
of NadR alone or in the presence of 4-HPA. MALLS analyses were
performed online with
SE-HPLC, using a Dawn TREOS MALLS detector (Wyatt Corp., Santa
Barbara, CA, USA)
and an incident laser wavelength of 658 nm. The intensity of the
scattered light was
measured at 3 angles simultaneously. Data elaboration was
performed using the Astra V
software (Wyatt) to determine the weighted-average absolute
molecular mass (MW), the
polydispersity index (MW/Mn) and homogeneity (Mz/Mn) for each
oligomer present in
solution. Normalization of the MALLS detectors was performed in
each analytical session by
use of bovine serum albumin.
Differential Scanning Calorimetry (DSC). The thermal stability
of NadR proteins
was assessed by DSC using a MicroCal VP-Capillary DSC instrument
(GE Healthcare).
NadR samples were prepared at a protein concentration of
0.5mg/mL (~30M) in buffer
containing 20mM Hepes, 300mM NaCl, pH 7.4, with or without 6mM
HPA or salicylate. The
DSC temperature scan ranged from 10°C to 110°C, with a thermal
ramping rate of 200°C per
hour and a 4 second filter period. Data were analyzed by
subtraction of the reference data for
a sample containing buffer only, using the Origin 7 software.
All experiments were performed
in duplicate, and mean values of the melting temperature (Tm)
were determined.
Surface plasmon resonance (SPR). Determination of equilibrium
dissociation
constant, KD: Surface plasmon resonance binding analyses were
performed using a Biacore
T200 instrument (GE Healthcare) equilibrated at 25 °C. The
ligand (NadR) was covalently
immobilized by amine-coupling on a CM-5 sensor chip (GE
Healthcare), using 20 µg/mL
purified protein in 10 mM sodium acetate buffer pH 5, injected
at 10 µl/min for 120 s until
~9000 response units (RU) were captured. A high level of ligand
immobilization was required
due to the small size of the analytes. An unmodified surface was
used as the reference
channel. Titrations with analytes (HPAs or salicylate) were
performed with a flow-rate of 30
µl/min, injecting the compounds in a concentration range of 10
µM to 20 mM, using filtered
running buffer containing Phosphate Buffered Saline (PBS) with
0.05 % Tween-20, pH 7.4.
Following each injection, sensor chip surfaces were regenerated
with a 30-second injection
of 10 mM Glycine pH 2.5. Each titration series contained 20
analyte injections and was
performed in triplicate. Titration experiments with long
injection phases (> 15 mins) were
used to enable steady-state analyses. Data were analyzed using
the BIAcore T200
evaluation software and the steady-state affinity model. A
buffer injection was subtracted
from each curve, and reference sensorgrams were subtracted from
experimental
sensorgrams to yield curves representing specific binding. The
equilibrium dissociation
constant, KD, was determined from the plot of RUeq against
analyte concentration, as
described previously [80].
-
24
Determination of binding stoichiometry: From each plot of RUeq
against analyte
concentration, obtained from triplicate experiments, the Rmax
value (maximum analyte binding
capacity of the surface) was extrapolated from the experimental.
Stoichiometry was
calculated using the molecular weight of dimeric NadR as ligand
molecule (MW ligand) and the
molecular weights of the HPA analyte molecules (MWanalyte), and
the following equation:
𝐒𝐭𝐨𝐢𝐜𝐡𝐢𝐨𝐦𝐞𝐭𝐫𝐲 =𝐑 𝑚𝑎𝑥 × 𝐌𝐖 𝑙𝑖𝑔𝑎𝑛𝑑
𝐌𝐖 𝑎𝑛𝑎𝑙𝑦𝑡𝑒 × 𝐑 𝑙𝑖𝑔𝑎𝑛𝑑
where Rligand is recorded directly from the sensorgram during
ligand immobilization prior to the
titration series, as described previously [81]). The
stoichiometry derived therefore
represented the number of HPA molecules bound to one dimeric
NadR protein.
Crystallization of NadR in the presence or absence of 4-HPA.
Purified NadR was
concentrated to 2.7 mg/mL using a centrifugal concentration
device (Amicon Ultra-15
Centrifugal Filter Unit with Ultracel-10 membrane with cut-off
size 10kDa; Millipore) running
at 600 g in a bench top centrifuge (Thermo Scientific IEC CL40R)
refrigerated at 2-8°C. More
concentrated samples were found to induce precipitation of the
protein. To prepare holo-
NadR samples, HPA ligands were added at a 200-fold molar excess
prior to the centrifugal
concentration step, and this ratio of protein:ligand
concentration was maintained. The
concentrated holo- or apo-NadR was subjected to crystallization
trials performed in 96-well
low-profile Intelli-Plates (Art Robbins) or 96-well low-profile
Greiner crystallization plates,
using a nanodroplet sitting-drop vapour-diffusion format and
mixing equal volumes (200nL) of
protein samples and crystallization buffers using a Gryphon
robot (Art Robbins).
Crystallization trays were incubated at 20º C. Crystals of
apo-NadR were obtained at 20ºC in
50 % PEG 3350 and 0.13 M di-Ammonium hydrogen citrate, whereas
crystals of SeMet–
NadR in complex with 4-HPA grew at 20ºC in condition H4 of the
Morpheus screen
(Molecular Dimensions), which contains 37.5% of the pre-mixed
precipitant stock
MPD_P1K_PEG 3350, buffer system 1 and 0.1 M amino acids, at a pH
6.5. All crystals were
mounted in cryo-loops using 10% ethylene glycol or 10% glycerol
as cryo-protectant before
cooling to 100 K for data collection.
X-ray diffraction data collection and structure determination.
X-ray diffraction
data from crystals of apo-NadR and SeMet–NadR/4-HPA were
collected on beamline PXII-
X10SA of the Swiss Light Source (SLS) at the Paul Scherrer
Institut (PSI), Villigen,
Switzerland. All diffraction data were processed in-house with
XDS [82] and programs from
the CCP4 suite [83]. Crystals of apo-NadR and 4-HPA-bound
SeMet-NadR belonged to
-
25
space group P 43 21 2 (see Table 2). Apo-NadR crystals contained
four protein molecules in
the asymmetric unit (Matthews coefficient 2.25 Å3·Da−1, for a
solvent content of 45 %), while
crystals of SeMet–NadR/4-HPA contained two protein molecules in
the asymmetric unit
(Matthews coefficient 1.98 Å3·Da−1, for a solvent content of 38
%). In solving the holo-NadR
structure, an initial and marginal molecular replacement (MR)
solution was obtained using as
template search model the crystal structure of the
transcriptional regulator PA4135 (PBD
entry 2FBI), the closest structurally-characterized homologue,
with which NadR shares ~54%
sequence identity. This solution was combined with SAD data to
aid identification of two
selenium sites in NadR, using autosol in phenix [84] and this
allowed generation of high-
quality electron density maps that were used to build and refine
the structure of the complex.
Electron densities were clearly observed for almost the entire
dimeric holo-NadR protein,
except for residues 88-90 of chain B, which lie in an exposed
region of the winged-helix motif
often found to be disordered in MarR family structures. The
crystal structure of apo-NadR
was subsequently solved by MR in Phaser [85] at 2.7 Å, using the
final refined model of
SeMet-NadR/4-HPA as the search model. For apo-NadR, electron
densities were clearly
observed for almost the entire protein, although residues 84-91
of chains A, C, and D, and
residues 84-90 of chain B lacked densities suggesting local
disorder or flexibility. Both
structures were refined and rebuilt using phenix [84] and Coot
[86], and structural validation
was performed using Molprobity [87]. Data collection and
refinement statistics are reported in
Table 2. Atomic coordinates of the two NadR structures have been
deposited in the Protein
Data Bank, with entry codes 5aip (NadR bound to 4-HPA) and 5aiq
(apo-NadR). All
crystallographic software was remotely compiled, installed and
maintained by SBGrid [88].
-
26
Results
NadR is dimeric and is stabilized by specific
hydroxyphenylacetate ligands.
Recombinant NadR was produced in E. coli using an expression
construct prepared
from the nadR gene of the N. meningitidis serogroup B strain
MC58. Standard
chromatographic techniques were used to purify NadR (see
Materials and Methods). In
analytical size-exclusion high-performance liquid chromatography
(SE-HPLC) experiments
coupled with multi-angle laser light scattering (MALLS)
analyses, NadR presented a single
species ≥ 97% pure with an absolute molecular mass of 35 kDa
(Figure 1.1)
Figure 1.1. Size-exclusion high-performance liquid
chromatography profile. SE-HPLC was used to assess the purity and
the apparent molecular weight of the recombinant wild-type (WT)
NadR sample alone (panel A) or containing a 200-fold molar excess
of 4-HPA (panel B). The NadR
monomer concentration was approximately 50M, and the 4-HPA
concentration was approximately 10mM. Data are plotted as
Absorbance Units (mAU) at 280nm wavelength, against retention time
in minutes. The elution time for the peak at maximum absorbance is
indicated in each panel. Analysis by integration of the peaks (peak
boundaries were defined as indicated by the red triangles),
revealed that the NadR sample was ≥ 97% pure. In both cases, the
protein eluted with a retention time (~ 22.5 minutes) indicative of
a dimer (~ 35kDa), determined by calibration of the column using
standard molecular weight markers (Bio-rad, cat. no. 151-1901).
Notably, the retention time was not significantly changed by the
presence of the ligand.
These data showed that NadR was dimeric in solution, since the
theoretical molecular
mass of the NadR dimer is 33.73 kDa. Subsequently, SE-HPLC/MALLS
analyses of NadR
A
B
NadR WT 17,5
AU (215nm)
0,00
0,20
0,40
0,60
0,80
1,00
1,20
1,40
1,60
Minutes 0 2 4 6 8 10 12 14 16 18 20
NadR WT + 4HPA
17,5
26,911
AU (215nm)
0,00
0,50
1,00
1,50
2,00
2,50
3,00
Minutes 0 2 4 6 8 10 12 14 16 18 20
-
27
were performed in the presence 4-HPA, revealing that there were
no changes in oligomeric
state upon addition of the ligand (Figure 1.2).
Figure 1.2. Multi-angle laser light scattering. MALLS analyses
were performed in order to determine the absolute molecular mass of
NadR alone (panel A) or in the presence of 4-HPA (panel B). The
curves plotted correspond to Absorbance Units (mAU) at 280nm
wavelength (green), light scattering (red), and refractive index
(blue). The elution peak maxima were at 17.5 minutes and the
numerical data obtained for absolute molecular mass and
polydispersity are shown below each image. In both cases, the MALLS
data clearly indicated a single monodisperse species of absolute
molecular mass ~ 37.5 kDa, corresponding to the dimeric form of
NadR. (The numbers ‘1’ at the bottom of the gradient-shaded slice
identify the beginning and end of each fraction-1, used for the
MALLS analyses
The thermal stability of NadR was examined using differential
scanning calorimetry
(DSC). Since ligand-binding can increase protein stability [89],
it was also investigated the
effect of various HPAs on the melting temperature (Tm) of NadR .
As a control of specificity it
was tested a salicylate, a ligand of MarR proteins reported to
increase the Tm of ST1710 and
MTH313 by approximately 3°C and 9°C, respectively [74]. In the
absence of ligand, the Tm of
apo-NadR was 67.3 ± 0.1°C. An increased thermal stability was
induced by 4-HPA (ΔTm
~ 3°C) and, to a lesser extent, by 3-HPA (ΔTm ~ 2°C) (Figure
1.3). Interestingly, NadR
displayed the greatest increase in thermal stability upon
addition of 3Cl,4-HPA (ΔTm ~ 4°C)
and was unaffected by salicylate.
-
28
Figure 1.3. Binding and thermostabilization of NadR by small
molecule ligands. (A) Molecular structures of salicylate and the
hydroxyphenylacetates tested. (B) DSC profiles are colored as
follows: apo-NadR (pink), NadR+salicylate (blue), NadR+3-HPA
(orange), NadR+4-HPA (green), NadR+3Cl,4-HPA (brown).
NadR displays distinct binding affinities for
hydroxyphenylacetate ligands
To further investigate the binding of HPAs to NadR, surface
plasmon resonance
(SPR) was used. The SPR sensorgrams displayed very fast
association and dissociation
events typical of small molecule ligands, thus prohibiting a
detailed kinetic study. However,
steady-state SPR analyses of the NadR-HPA interactions readily
allowed determination of
the equilibrium dissociation constants (KD). KD values of
interaction of ligands with NadR are
reported in Table 1.1, showing that 3Cl,4-HPA was the tightest
binder, and thus matched the
ranking of ligand-induced Tm increases observed in the DSC
experiments. Although these KD
values indicate relatively weak interactions, they are similar
to the values determined for the
MarR/salicylate interaction (KD~1mM) [90] and the
MTH313/salicylate interaction (KD 2-3mM)
[74], and are approximately 20-fold tighter than the
ST1710/salicylate interaction (KD ~20mM)
[75].
Ligand ∆Tm (°C) KD (mM)
Salicylate 0 -
3-HPA 2.7 2.7 ± 0.1
4-HPA 3.3 1.5 ± 0.1
3Cl,4-HPA 3.9 1.1 ± 0.1
Table 1.1. Thermal stabilization (∆Tm) and dissociation
constants (KD) of the NadR/ligand interactions.
3 - HPA
3,4 - HPA Salicylate
4 - HPA
3Cl - 4 - HPA
C l
A B
50
60
70
80
90
0
2
4
6
8
10
Cp
(kcal/m
ole
/ o
C)
Temperature ( o C)
No ligand
3Cl,4-HPA
3-HPA
Salicylate
∆Tm
4-HPA
-
29
Crystal structures of holo-NadR and apo-NadR
To fully characterize the NadR/HPA interactions, the crystal
structures of ligand-
bound (holo) and ligand-free (apo) NadR was determined. First,
NadR was crystallized (a
selenomethionine (SeMet)-labelled form) in the presence of a
200-fold molar excess of 4-
HPA. The structure of the NadR/4-HPA complex was determined at
2.3 Å resolution using a
combination of the single-wavelength anomalous dispersion (SAD)
and molecular
replacement (MR) methods, and was refined to Rwork/Rfree values
of 20.9/26.0 % (Table 1.2).
NadR SeMet + 4-HPA (SAD peak) (PDB code 5aip)
NadR apo-form (PDB code 5aiq)
Data collection
Wavelength (Å) 0.9792 1.0 Beamline SLS (PXII-X10SA) SLS
(PXII-X10SA)
Resolution range (Å) 39.2 - 2.3 48.2 - 2.7 Space group P 43 21 2
P 43 21 2
Unit cell 75.3, 75.3, 91.8 69.4, 69.4, 253.8
Total reflections 291132 (41090) 225521 (35809)
Unique reflections 12320 (1773) 17700 (2780) Multiplicity 23.6
(23.2) 12.7 (12.8)
Completeness (%) 100.0 (100.00) 99.9 (99.7)
Mean I/sigma(I) 25.5 (9.0) 22.6 (3.8) Wilson B-factor 23.9
49.1
Rsym* 10.9 (39.4) 11.4 (77.6)
Rmeas** 11.3 11.8
Refinement
Rwork♯ 20.9 21.7
Rfree♯♯ 26.0 27.2
Number of atoms
Non-hydrogen atoms
2263
4163
Macromolecules 2207 4144 Ligands 11 0
Water 45 19
Protein residues 275 521 RMS(bonds) 0.008 0.003
RMS(angles) 1.09 0.823
Ramachandran (%)§ Favored 100 98.4
Outliers 0 0
Clashscore 5.0 3.9 Average B-factor
Macromolecules 34.8 53.3 Ligands 32.9 -
Solvent 37.3 (H2O) 29.0 (H2O)
Statistics for the highest-resolution shell are shown in
parentheses. * Rsym = Σhkl Σi |Ii(hkl) - | / Σhkl Σi Ii(hkl) **
Rmeas = redundancy-independent (multiplicity-weighted) Rmerge as
reported from AIMLESS [91]. ♯ Rwork = Σ||F(obs)|-
|F(calc)||/Σ|F(obs)| ♯♯ Rfree = as for Rwork, calculated for 5.0%
of the total reflections, chosen at random, and omitted from
refinement. § Figures from Molprobity [87].
Table 1.2. Data collection and refinement statistics for NadR
structures.
X-ray crystallography was selected as the method-of-choice, due
to its well-known
capacity to provide high-resolution information about
protein-small molecule interactions.
-
30
NMR spectroscopy was a possible alternative structural
technique, but since a number of
MarR family proteins had been previously crystallized, it was
considered likely that the
crystallographic approach would have a reasonable
probability-of-success. In contrast, the
NadR protein dimer was considered too small to be tractable by
the recently-emerging
electron cryomicroscopy techniques, which are better suited for
larger macromolecules [92]
Despite numerous attempts, it was not possible to obtain
high-quality crystals of
NadR complexed with 3Cl,4-HPA, 3,4-HPA, 3-HPA or DNA targets.
However, it was possible
to crystallize apo-NadR, and the structure was determined at 2.7
Å resolution by MR using
the NadR/4-HPA complex as the search model. The apo-NadR
structure was refined to
Rwork/Rfree values of 19.1/26.8 % (Table 1.2).
The asymmetric unit of the NadR/4-HPA crystals (holo-NadR)
contained one NadR
homodimer, while the apo-NadR crystals contained two homodimers.
In the apo-NadR
crystals, the two homodimers are related by a rotation of ~90º;
the observed association of
the two dimers was presumably an effect of crystal packing,
since the interface between the
two homodimers is small (< 550 Å2 of buried surface area),
and is not predicted to be
physiologically relevant by the PISA software [93]. Moreover,
our SE-HPLC/MALLS analyses
revealed that in solution NadR is dimeric, and previous studies
using native mass
spectrometry (MS) also revealed dimers and not tetramers
[94].
The holo-NadR homodimer shows a dimerization interface mostly
involving the top of
its triangular form, while the two DNA-binding domains are
located at the base The overall
structure of NadR shows triangular dimensions of ~50 × 65 × 50 Å
and a large homodimer
interface burying a total surface area ~ 4800 Å2 (Figure
1.4).
Figure 1.4. The crystal structure of NadR in complex with 4-HPA.
(A) The holo NadR homodimer is depicted in green and blue for
chains A and B respectively, while yellow sticks depict the 4-HPA
ligand. Secondary structures are labelled for chain B only. (B)
Orientation as in panel A, showing the secondary structure elements
of NadR protein. Red dashes in panels A and B show hypothetical
positions of chain B residues 88-90 that were not modeled due to
lack of electron density.
-
31
Each NadR monomer consists of six α-helices and two short
β-strands, with helices
α1, α5, and α6 forming the dimerization interface. Helices α3
and α4 form a helix-turn-helix
motif, followed by the “wing motif” comprised of two short
antiparallel β-strands (β1-β2).
These secondary structure elements constitute the winged
helix-turn-helix (wHTH) DNA-
binding domain and, together with the dimeric organization, are
the hallmarks of MarR family
structures [72].
The holo-NadR structure presents only one occupied
ligand-binding pocket
As already shown in Figure1.4, the NadR/4-HPA structure revealed
the ligand-binding
site nestled between the dimerization and DNA-binding domains.
High-quality electron
density maps allowed clear identification of the bound 4-HPA
ligand, which showed a
different position and orientation compared to salicylate
complexed with MTH313 and
ST1710 [74, 75] (see Discussion). The binding pocket was almost
entirely filled by 4-HPA
and one water molecule, although there also remained a small
tunnel 2-4Å in diameter and
5-6Å long leading from the pocket (proximal to the 4-hydroxyl
position) to the protein surface.
The tunnel was lined with rather hydrophobic amino acids, and
did not contain water
molecules. Most unexpectedly, only one monomer of the NadR
homodimer contained 4-HPA
in the binding pocket, whereas the corresponding pocket position
of the other monomer was
unoccupied by ligand.
Inspection of the protein-ligand interaction network revealed no
bonds from NadR
backbone NH or CO groups to the ligand, but several key side
chain mediated hydrogen (H)-
bonds and ionic interactions, most notably between the
carboxylate group of 4-HPA and
Ser9 (chain A), and Trp39, Arg43 and Tyr115 of chain B (Figure
1.5A). At the other end of
the ligand, the 4-hydroxyl group was proximal to H-bond donors
in the side chains of Asn11
(chain A) and Asp36 (chain B), although these were positioned at
slightly greater distances
(3.6-4.3 Å) than those atoms contacting the carboxylate group.
There was also one water
molecule observed in the pocket, bound by the carboxylate group
and the side chains of
Ser9 and Asn11 from chain A.
-
32
Figure 1.5. Atomic details of the NadR/4-HPA interaction. (A)
View of the binding pocket showing side chain sticks for all
interactions between NadR and 4-HPA. Green and blue ribbons depict
NadR chains A and B, respectively. 4-HPA is shown in yellow sticks,
with oxygen atoms in red. A water molecule is shown by the red
sphere. A list of the interacting atoms and bond distances is
provided in Table 3. Side chains mediating hydrophobic interactions
are shown in orange. The yellow sphere on the 4-HPA phenyl ring
shows the 3-position at which the chloro group of 3Cl,4-HPA could
be readily accommodated. B) 4-HPA is sandwiched by NadR, as shown
by the surface representation of residues that line the binding
pocket. ‘Ceiling’ residues are colored orange, ‘floor’ residues are
colored blue (chain B) or green (chain A).
In addition to the H-bonds involving the carboxylate and
hydroxyl groups of 4-HPA,
binding of the phenyl moiety appeared to be stabilized by van
der Waals’ interactions
involving the hydrophobic side chain atoms of Arg18 (via the Cβ,
Cγ, Cδ methylene groups),
Leu21, Met22, Phe25, Leu29 and Val111 of chain B (Figure 1.5A).
In particular, the phenyl
ring of Phe25 was positioned parallel to the phenyl ring of
4-HPA, potentially forming -
parallel-displaced stacking interactions. Interestingly, NadR
residues in the 4-HPA binding
pocket effectively created a polar ‘floor’ and a hydrophobic
‘ceiling’, which house the ligand
(Figure 1.5B). The polar floor is made of residues both from
chain A and chain B of the
homodimer, while the ceiling is made of residues from chain B
only (Figure 1.5B).
Collectively, this mixed network of polar and hydrophobic
interactions endows NadR with a
strong recognition pattern for HPAs, with additional
medium-range interactions potentially
established to the hydroxyl group at the 4-position.
Analysis of the pockets reveals the molecular basis for
asymmetry and stoichiometry
The lack of a second 4-HPA molecule in the homodimer suggested
negative co-
operativity, a phenomenon previously described for the
MTH313/salicylate interaction [74]
and for other MarR family proteins [72]. To understand the
molecular basis of asymmetry in
NadR, the ligand-free monomer (chain A) was superposed onto the
ligand-occupied
monomer (chain B). Overall, the superposition revealed a high
degree of structural similarity
A B
-
33
(Cα root mean square deviation (rmsd) of 1.5Å), though on closer
inspection a rotational
difference of ~9 degrees of helix α6 was observed, suggesting
that 4-HPA induced a slight
conformational change (Figure 1.6A)
Figure 1.6. Structural differences between the two monomers of
holo-NadR. (A) Aligned monomers of the holo-NadR structure,
revealing that the major overall difference is the ~9 degree shift
in the position of helix α6 (chain A: green; chain B: blue). (B) A
closer comparison of the binding pockets shows that in the
ligand-free monomer chain A (green) residues M22, F25 and R43 adopt
‘inward’ positions (highlighted by arrows) that would prevent
binding of 4-HPA due to clashes with the 4-hydroxyl group, the
phenyl ring and the carboxylate group, respectively. (Both panels
(A) and (B) are rotated compared to Figure 1.5).
However, since residues of helix α6 were not directly involved
in ligand binding, an
explanation for the lack of 4-HPA in monomer A did not emerge by
analyzing only the
backbone atom positions suggesting that a more complex series of
allosteric events may
occur. Indeed, it was noted interesting differences in the side
chains of Met22, Phe25 and
Arg43, which in monomer B are used to contact the ligand while
in monomer A they partially
occupied the pocket and collectively reduced its volume
significantly. Specifically, upon
analysis with the CASTp software [95], the pocket in chain B
containing the 4-HPA exhibited
a total volume of 368Å3, while the pocket in chain A was
occupied by side chains and was
divided into three much smaller pockets, each with volumes <
50Å3, evidently rendering
chain A unfavorable for ligand binding. Most notably, atomic
clashes between the ligand and
the side chains of Met22, Phe25 and Arg43 (chain A) would occur
if 4-HPA were present in
the monomer A pocket (Figure 1.6B). Subsequently, analyses of
the pockets in apo-NadR
revealed that in the absence of ligand the long Arg43 side chain
was always in the open
‘outward’ position compatible with binding to the 4-HPA
carboxylate group. In contrast, the
apo-form Met22 and Phe25 residues were still encroaching the
spaces of the 4-hydroxyl
group and the phenyl ring of the ligand, respectively (Figure
1.6B). The ‘outward’ position of
A B
-
34
Arg43 generated an open apo-form pocket with volume
approximately 380Å3. Taken
together, these observations suggest that Arg43 is a major
determinant of ligand binding,
and that its ‘inward’ position inhibits the binding of 4-HPA to
the empty pocket of holo-NadR.
To support the crystallographic data, the binding stoichiometry
was investigated using
solution-based techniques. However, studies based on tryptophan
fluorescence were
confounded by the fluorescence of the HPA ligands, and
isothermal titration calorimetry (ITC)
was unfeasible due to the need for very high concentrations of
NadR in the ITC chamber
(due to the relatively low affinity), which exceeded the
solubility limits of the protein.
However, it was possible to calculate the binding stoichiometry
of the NadR-HPA interactions
using an SPR-based approach. In SPR, the signal measured is
proportional to the total
molecular mass proximal to the sensor surface; consequently, if
the molecular weights of the
interactors are known, then the stoichiometry of the resulting
complex can be determined
[81]. This approach relies on the assumption that the captured
protein (‘the ligand’, according
to SPR conventions) is 100 % active and freely-accessible to
potential interactors (‘the
analytes’). This assumption is likely valid for this pair of
interactors, for two main reasons.
Firstly, NadR is expected to be covalently immobilized on the
sensor chip as a dimer in
random orientations, since it is a stable dimer in solution and
has sixteen lysines well-
distributed around its surface, all able to act as potential
sites for amine coupling to the chip,
and none of which are close to the ligand-binding pocket.
Secondly, the HPA analytes are all
very small (MW 150-170) and therefore are expected to be able to
diffuse readily into all
potential binding sites, irrespective of the random orientations
of the immobilized NadR
dimers on the chip. The stoichiometry of the NadR-HPA
interactions was determined using
Equation 1 (see Materials and Methods), and revealed
stoichiometries of 1.13 for 4-HPA,
1.02 for 3-HPA, and 1.21 for 3Cl,4-HPA, strongly suggesting that
one NadR dimer bound to 1
HPA analyte molecule.
Apo-NadR structures reveal conformational flexibility
After determination of the holo-NadR structure, the structure of
apo-NadR was
determined. The apo-NadR structure contained two homodimers in
the asymmetric unit
(chains A+B and chains C+D), which upon superposition revealed a
few minor differences
and an rmsd of 1.55Å. Similarly, superpositions of the
holo-homodimer onto each of the apo-
homodimers resulted in rmsd values of 1.29Å and 1.31Å, again
showing some slight overall
differences between the homodimer pairs. The slightly larger
difference between the two
apo-homodimers, rather than between apo- and holo-homodimers,
indicated that apo-NadR
possesses a notable degree of conformational flexibility. The
overall structural similarity but
-
35
with inherent plasticity of MarR proteins was observed
previously upon comparison of the
OhrR, MarR, MexR and SarR structures [96].
4-HPA stabilizes concerted conformational changes in NadR that
prevent DNA-binding
To further investigate the conformational rearrangements of
NadR, local structural
alignments were performed using a subset of residues in the
DNA-binding helix. By selecting
and aligning residues Arg64-Ala77 of one α4 helix from each
homodimer, superposition of
the holo-homodimer onto the two apo-homodimers revealed
differences in the monomer
conformations of each structure (Figure1.7A). While one monomer
from each structure was
closely superimposable (compare green and cyan cartoons,
Figure1.7A), the second
monomer displayed quite large differences, especially in the
DNA-binding helix 4 which
shifted by as much as 6Å (Figure1.7B). Accordingly, helix α4 was
also found to be one of the
most dynamic regions of NadR in previous HDX-MS analyses
[94].
Figure 1.7. Structural comparison of holo- and apo-NadR and
modelling of interactions with DNA. (A) The holo-homodimer
structure is shown as green and blue cartoons, for chain A and B,
respectively, while the two homodimers of apo-NadR are both cyan
and pale blue for chains A and B, respectively. The three
homodimers (chains AB holo, AB apo, and CD apo) were overlaid by
structural alignment of all heavy atoms in residues R64-A77 (shown
in red, with side chain sticks) of chains A holo, A apo, and C apo,
belonging to helix α4 (left). The α4 helices aligned closely, Cα
rmsd 0.2Å for 14 residues. (B) The relative positions of the α4
helices of the 4-HPA-bound holo homodimer chain B (blue), and of
apo homodimers AB and CD (showing chains B and D) in pale blue.
Dashes indicate the Ala77 Cα atoms, in the most highly shifted
region of the ‘non-fixed’ α4 helix.
However, structural comparisons revealed that the shift of
holo-NadR helix α4
induced by the presence of 4-HPA was also accompanied by several
changes at the holo
dimer interface, while such extensive structural differences
were not observed in the apo
dimer interfaces, particularly notable when comparing the α6
helices (Figure1.7A). In
summary, compared to ligand-stabilized holo-NadR, apo-NadR
displayed an intrinsic
A B
-
36
flexibility focused in the DNA-binding region. This was also
evident in the greater disorder (i.e.
less well-defined electron density) in the β1-β2 loops of the
apo dimers (density for 16
residues per dimer was missing) compared to the holo dimer
(density for only 3 residues was
missing).
In holo-NadR, the distance separating the two DNA-binding α4
helices was 32 Å,
while in apo-NadR it was 29 Å for homodimer AB, and 34 Å for
homodimer CD. Thus, the
apo-homodimer AB presented the DNA-binding helices in a
conformation similar to that
observed in the protein:DNA complex OhrR:ohrA from Bacillus
subtilis [96] (Figure1.8A).
Figure 1.8. Structural comparison of holo- and apo-NadR and
modelling of interactions with DNA. (A) The holo- and the apo- NadR
homodimer structures are shown and superimposed as already reported
in Figure 1.7. (A) The double-stranded DNA molecule (grey cartoon)
from the OhrR-ohrA complex is shown after superposition with NadR,
to highlight the expected positions of the NadR α4 helices in the
B-DNA major grooves. The proteins share ~30% amino acid sequence
identity. For clarity, only the α4 helices are shown in panels (A)
and (B). (B) Upon comparison with the experimentally-determined
OhrR:ohrA structure (grey), the α4 helix of holo-NadR (blue) is
shifted ~8Å out of the major groove.
Interestingly, OhrR contacted ohrA across 22 base pairs (bp),
and similarly the main
NadR target sites identified in the nadA promoter (the operators
Op I and Op II) were both
shown to span 22 bp [63, 64]. Pairwise superpositions showed
that the NadR apo-
homodimer AB was the most similar to OhrR (rmsd 2.6Å), while the
holo-homodimer was the
most divergent (rmsd 3.3Å) (Figure1.7A). Assuming the same
overall DNA-binding
mechanism is used by OhrR and NadR, the apo-homodimer AB was
ideally pre-configured
for DNA binding, while 4-HPA appeared to stabilize holo-NadR in
a conformation poorly
suited for DNA binding. When aligned with OhrR, the
apo-homodimer CD presented another
different intermediate conformation (rmsd 2.9Å), apparently not
ideally pre-configured for
DNA binding, but which in solution can presumably readily adopt
the AB conformation due to
the intrinsic flexibility described above. In addition to the
different inter-helical translational
distances, the 4 helices in the holo-NadR homodimer had also
rotated, resulting in
movement of α4 out of the major groove and preventing efficient
DNA binding in the
presence of 4-HPA (Figure1.8B).
A B
-
37
A single conserved leucine residue (L130) is crucial for
dimerization
To study the architecture and stability of the NadR homodimer
interface, a series of
mutations were prepared, with the aim of disrupting the dimer
interface. Due to the two-fold
symmetry of the interface, each amino acid exchange disrupts
twice a given dimer contact.
The NadR dimer interface is formed by at least 32 residues,
which establish
numerous inter-chain salt bridges or hydrogen bonds, and many
hydrophobic packing
interactions (Figure1.9 A and B). To determine which residues
were most important for
dimerization, the interface in silico was studied and several
residues were identified as
potential mediators of key stabilizing interactions. Using
site-directed mutagenesis, a panel of
eight mutant NadR proteins was prepared (including mutations
H7A, S9A, N11A, D112A,
R114A, Y115A, K126A, L130K and L133K), sufficient to explore the
entire dimer interface.
The crystal structures presented here allowed a detailed
structural analysis for the
design of NadR mutants:
H7, S9, N11 are relevant residues establishing hydrogen bonds
interactions both in
4HPA binding pocket;
Y115 is main-chain interaction with N11 and side-chain with S9,
removing it could
abolish both potential contacts, between perpendicular
helices;
location of the residue K126 suggests that this might be the
main contributor to the
interface, being on helix α6 and symmetrical;
K130 is located on helix α6 and mutated in Lys to introduce a
long charged residue
in place of hydrophobic residues;
A B
-
38
Figure 1.9 Analysis of the NadR dimer interface. (A) Both
orientations show chain A, green backbone ribbon, colored red to
highlight all locations involved in dimerization; namely,
inter-chain salt bridges or hydrogen bonds involving Q4, S5, K6,
H7, S9, I10, N11, I15, Q16, R18, D36, R43, A46, Q59, C61, Y104,
D112, R114, Y115, D116, E119, K126, E136, E141, N145, and the
hydrophobic packing interactions involving I10, I12, L14, I15, R18,
Y115, I118, L130, L133, L134 and L137. Chain B, grey surface, is
marked blue to highlight residues probed by site-directed
mutagenesis (E136 only makes a salt bridge with K126, therefore it
was sufficient to make the K126A mutation to assess the importance
of this ionic interaction; the H7 position is labelled for monomer
A, since electron density was lacking for monomer B). (B) A zoom
into the environment of helix α6 to show how residue L130 chain B
(blue side chain) is a focus of hydrophobic packing interactions
with L130, L133, L134 and L137 of chain A (red side chains). (C)
SE-HPLC analyses of all mutant forms of NadR are compared with the
wild-type (WT) protein. The WT and most of the mutants show a
single elution peak with an absorbance maximum at 17.5 min. Only
the mutation L130K has a noteworthy effect on the oligomeric state,
inducing a second peak with a longer retention time and a second
peak maximum at 18.6 min. To a much lesser extent, the L133K
mutation also appears to induce a ‘shoulder’ to the main peak,
suggesting very weak ability to disrupt the dimer. (D)
SE-HPLC/MALLS analyses of the L130K mutant, shows 20% dimer and 80%
monomer. The curves plotted correspond to Absorbance Units (mAU)
at